Jessie A Ellis
22 Jul 2024 16:37
According to a Lisbon Council study report, AI and accelerated computing are increasing energy efficiency across a wide range of sectors and significantly reducing consumption and carbon emissions.
According to a report from Lisbon Council Research, artificial intelligence (AI) and accelerated computing are making significant progress in improving energy efficiency across a number of industries. These advances are crucial as data centers are expected to account for up to 4% of global energy consumption in the near future.
Why Accelerated Computing is Sustainable Computing
Accelerated computing leverages the parallel processing capabilities of NVIDIA GPUs to complete more work in less time and consume less energy than traditional CPU-based systems. This form of computing is increasingly recognized as sustainable computing.
Given that AI is inherently parallel, integrating AI amplifies the benefits. A report from Lisbon Council Research highlights that AI applications such as machine learning and deep learning perform much better on GPUs than on CPUs, resulting in significant energy savings.
User Experience with Accelerated AI
Industries around the world are reporting remarkable energy efficiency gains using AI and accelerated computing. For example, Murex, a financial services company, tested NVIDIA’s Grace Hopper Superchip and achieved a 4x reduction in energy consumption and a 7x reduction in completion time compared to a CPU-only system.
Similarly, Taiwan-based Wistron used NVIDIA Omniverse to create a digital twin of its test room, improving energy efficiency by up to 10 percent, resulting in a 120,000 kWh reduction in annual electricity consumption and a 60,000 kg reduction in carbon emissions.
Reduce carbon emissions by up to 80%
The RAPIDS Accelerator for Apache Spark has demonstrated the potential to reduce the carbon footprint of data analytics by up to 80%, with average speedups of up to 5x and compute costs reduced by 4x. Major enterprises such as Adobe, AT&T, and the Internal Revenue Service are leveraging this technology.
In healthcare, Insilico Medicine leverages NVIDIA-powered AI to discover drug candidates for rare respiratory diseases that have reached Phase 2 clinical trials at a fraction of the typical cost and time.
Accelerating Science with Accelerated AI
The National Energy Research Scientific Computing Center (NERSC) also reported significant energy efficiency gains with the NVIDIA A100 Tensor Core GPUs. Their applications saw an average 5x increase in energy efficiency, and their weather forecasting application saw an improvement of nearly 10x.
A recent ranking of the world’s most energy-efficient supercomputers highlighted the significant energy savings that accelerated computing can deliver.
Underestimated Energy Savings
Despite these benefits, some predictions only consider the energy consumed during training AI models and overlook the efficiency gains during the deployment phase. A study from the Information Technology and Innovation Foundation (ITIF) refutes these predictions, highlighting the significant energy savings that AI can provide after training.
The report urges policymakers to recognize the potential of AI to achieve a low-carbon future and to support its adoption to maximise economic and social benefits.
AI supports sustainability efforts
The role of AI in promoting sustainability is further supported by a number of reports. For example, AI can improve weather modeling accuracy, improve crop yield forecasts, and help discover efficient battery materials.
Governments are being encouraged to adopt AI tools to decarbonize their operations, and NVIDIA is working with a number of startups to tackle climate challenges and develop Earth-2, an AI supercomputer dedicated to climate science.
Improved energy efficiency across the stack
NVIDIA has focused on improving energy efficiency across its products since its inception. Recent innovations include the NVIDIA GB200 Grace Blackwell Superchip, which delivers 25x more energy efficiency than the previous generation for AI inference, and the BlueField-3 DPU, which offloads data center functions from the CPU, reducing power consumption by up to 30%.
Moreover, NVIDIA’s liquid cooling technology, which is being developed with support from the U.S. Department of Energy, promises a 20 percent efficiency improvement over current air-cooled systems.
These developments highlight NVIDIA’s commitment to increasing energy efficiency and supporting sustainable computing practices across a wide range of industries.
For more details, please see the original source.
Image source: Shutterstock